|
| Big Data and Analytics in Behavior Analysis |
| Sunday, May 24, 2020 |
| 9:00 AM–10:50 AM |
| Virtual |
| Area: OBM; Domain: Service Delivery |
| Chair: Jonathan E. Friedel (National Institute for Occupational Safety and Health) |
| Discussant: Mark P. Alavosius (Praxis2LLC) |
| Abstract: In the current age of advanced computing power and big data, more and more companies are looking to data mining and predictive analytics to reveal actionable insights. The promise of analytics has already been demonstrated in fields such as supply chain logistics, sports, healthcare, and marketing. In many cases, especially those pertaining to consumer preferences and behavior, the insights gleaned from analytics ultimately are meant to improve the prediction and control of human behavior—a common goal of behavior analysts. This symposium highlights several examples of the use of data mining and analytics to aid decision makers involved in two applications: organizational behavior management and behavioral safety. Four presentations were assembled to (1) summarize existing examples of analytics in occupational safety and the promise of analytics in enhancing the utility of behavioral safety data collections systems, (2) identify the foundational and organizational requisites for conducting analytics, (3) assess and evaluate an organization’s readiness for analytics, (4) provide examples of analytics applications and methods. |
| |
| An Overview of Analytics in Occupational Safety Research |
| OLIVER WIRTH (CDC/NIOSH), Jonathan E. Friedel (National Institute for Occupational Safety and Health), Anne M. Foreman (CDC/NIOSH) |
| Abstract: Advances in computing power and analytical methods have allowed companies to reveal insights from the “big” data that previously would have gone undetected in areas such as supply chain management, healthcare spending, and marketing. Analytic strategies offer a great deal of potential in the area of occupational safety and health because large amounts of data are collected within behavioral safety programs, including peer observations, safety audits, and near miss reports, and these data can be combined with data from other company departments (e.g., production, human resources) to better predict and prevent injuries and fatalities. Despite its potential, the progress of analytics in occupational safety has lagged other industries. The presentation will review the analytics studies that have been conducted in occupational safety and health at the industry, enterprise, and establishment levels. The components of the studies that will be reviewed include the industries, data types, modeling approaches, and findings. The potential advantages of establishment-level analytics in organizations with robust behavioral safety programs will be discussed. |
| |
| Utilizing Data Analytics to Inform Safety Interventions and Reduce Adverse Safety Outcomes |
| MATTHEW M LASKE (Appalachian State University), Yalcin Acikgoz (Appalachian State University), Timothy D. Ludwig (Appalachian State University), Shawn Bergman (Appalachian State University) |
| Abstract: Data analytics is a becoming more relevant for OBM practitioners. In particular, there is a growing need for analytics in safety for the detection of behavioral precursors associated with injuries and other safety outcomes. Although workplace incidents have been decreasing over the last 25 years, the number of fatalities has remained more or less constant, with 6,217 in 1992 and 5,147 in 2017 (United States Department of Labor [USDL], Bureau of Labor Statistics [BLS], 2018). Analytics can be utilized to inform OBM interventions, particularly in the field of Behavior Based Safety. A case study will be presented on the identified behavioral covariates of leading and lagging safety variables within a Fortune 500 chemical manufacturer. Results relating to production, employee workloads (e.g., overtime hours and unplanned work), and behavioral observation processes will be discussed. The purpose of the analyses is to identify interventions for which safety practitioners can allocate resources to reduce the risk of adverse safety outcomes (e.g., injuries and fatalities). |
| |
| Assessing Analytics Readiness Within Occupational Health and Safety |
| MAIRA COMPAGNONE (Appalachian State University), Timothy D. Ludwig (Appalachian State University), Shawn Bergman (Appalachian State University), Yalcin Acikgoz (Appalachian State University) |
| Abstract: The safety industry is lagging in Big Data utilization due to various obstacles, which may include lack of analytics readiness (e.g. disparate databases, missing data, low validity) or knowledge (e.g. personnel capable of cleaning data and running analyses). A safety-specific maturity model can assist organizations with identifying these obstacles, so that organizations can understand their current capabilities and can build towards more advanced analytics (e.g. predicting safety incidents and identifying preventative measures directed towards specific risk variables). These analytics can inform organizational behavior management practitioners in designing more effective interventions in behavioral safety programs. The model will (a) evaluate the quality of the data currently available, (b) evaluate the foundational infrastructure for technological capabilities and expertise in data collection, storage, and analysis of safety and health metrics, (c) evaluate the culture around employee willingness to share safety issues and concerns (e.g. participate in reporting, audits, inspections, and observations) with frequency and fidelity, and (d) assess an organization’s safety analytics process for adequacy of operational definitions (e.g. key safety indicators of “lagging” outcome variables and pre-incident “leading” indicators) and hypothesized variable relationships. The proposed model will be validated against data from two manufacturing organizations. |
| |
| The Scope and Variety of Analytics: Examples of Techniques Suitable for Behavioral Data Sets |
| ANNE M. FOREMAN (CDC/NIOSH), Jonathan E. Friedel (National Institute for Occupational Safety and Health), Oliver Wirth (CDC/NIOSH) |
| Abstract: The application of analytics is ubiquitous in fields such as supply chain logistics, sports, healthcare, and marketing. The widespread application of analytics in behavior analysis has not yet been realized, but its potential to provide actionable insights are nevertheless promising. This presentation will provide a survey of the various kinds of analytical techniques that fall under the broad umbrella of analytics. Specific methods include data visualization, machine learning, classification trees, neural networks, and support vector machines, to name a few. Potential applications of each technique will be provided with examples taken from organizational behavior management, behavioral safety, and other applied behavioral analysis fields. |
|
| |